Yes. At the very least, it's a template in which to enter variables that are testable. Comparing to the distant past isn't particularly useful, because we have limited and spotty information from the past and because the same conditions are never exactly duplicated. In the case of anthropogenic situations (not only climate change), the conditions have never existed before. So, in this regard all large-scale predictions are untestable. But trends can spotted and dangers foreseen - and, at least in theory, avoided.